Research map / Probabilistic and scientific modeling / Molecules and proteins
Errors-in-variables Fr\'echet Regression with Low-rank Covariate Approximation
Dogyoon Song, Kyunghee Han
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply singular value thresholding (SVT, low-rank approximation) instead of naive EIV | global Frechet regression · Gaussian errors, p=150, n=100 | MSPE: 1.267 -> 0.686 | Improved |
| apply SVT instead of naive EIV | global Frechet regression · Gaussian errors, p=600, n=400 | MSPE: 1.242 -> 0.657 | Improved |
| apply SVT instead of naive EIV | global Frechet regression · Laplacian errors, p=300, n=200 | MSPE: 1.521 -> 0.677 | Improved |
| apply SVT instead of naive EIV, smallest gap | global Frechet regression · Gaussian errors, p=600, n=100 | MSPE: 1.062 -> 1.054 | Improved |
| use SVT on error-prone covariates instead of oracle REF on error-free covariates | global Frechet regression · Gaussian errors, p=150, n=100 | MSPE: 1.613 -> 0.686 | Improved |
| use naive EIV on error-prone covariates instead of oracle REF | global Frechet regression · Gaussian errors, p=150, n=100 | MSPE: 1.613 -> 1.267 | Improved |
| use oracle REF instead of SVT | global Frechet regression · Gaussian errors, p=150, n=100 | bias: 0.312 -> 0.107 | Improved |
| use SVT instead of REF on variance | global Frechet regression · Gaussian errors, p=150, n=100 | sqrt(Var): 1.147 -> 0.531 | Improved |
| use oracle REF instead of SVT on in-sample fit | global Frechet regression · Gaussian errors, p=300, n=100 | MSE: 0.217 -> 0.000 | Improved |
| increase covariate dimension p from 150 to 600 | SVT · Gaussian errors, n=100 | MSPE: 0.686 -> 1.054 | Hurt |
| increase sample size n from 100 to 400 | SVT · Gaussian errors, p=150 | MSPE: 0.686 -> 0.412 | Improved |
| change error distribution from Gaussian to Laplacian | naive EIV · p=150, n=100 | MSPE: 1.267 -> 1.538 | Hurt |
Closest papers by meaning
- On Robustness of Principal Component Regression (NeurIPS 2019 · similarity 0.87)
- Canonical Correlation Analysis as Reduced Rank Regression in High Dimensions (JMLR 2026 · similarity 0.85)
- No Double Descent in Principal Component Regression: A High-Dimensional Analysis (ICML 2024 · similarity 0.85)
- Reduced-Rank Tensor-on-Tensor Regression and Tensor-Variate Analysis of Variance (TPAMI 2022 · similarity 0.85)
- Multiple Output Regression with Latent Noise (JMLR 2016 · similarity 0.84)
- High-Dimensional Tensor Regression With Oracle Properties (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
